Are there more qualified unemployed people when the economy is in recession? Are the unemployed then more often forced to accept low-quality jobs? Conversely, do workers more easily find better jobs when the economy rebounds? In order to answer these questions, Jean-Marc Robin, a researcher in Sciences Po’s department of economics, and Jeremy Lise, associate professor at the University of Minnesota, developed a very rich macroeconomic model that they calibrated with American data. They presented it in the highly prestigious The American Economic Review: The Macro-dynamics of Sorting between Workers and Firms, April 2017. Snapshot.
To understand the effect of the macroeconomic environment on the distribution of unemployed people’s qualifications, the quality of employment found and labor reallocations, Jean-Marc Robin and Jeremy Lise developed a model over four years that includes an unprecedented range of behaviors. The idea was to represent a labor market composed of multiple heterogeneous agents subjected to aggregate productivity shocks. These shocks feed the economy’s cyclical dynamic, commonly called economic conditions.
First, they imagined a fixed population of workers with uneven skills and an uneven distribution of vacancies. As is the case in reality, the quantity and quality of vacancies varied depending on cost-benefit arbitrages specific to each type of job. The distribution of workers’ skills remained the same. The distribution of the quality of available jobs changed depending on the economic conditions.
Sorting and economic conditions
The model’s mechanism is such that when economic conditions are good, more jobs are created, the unemployed more easily find work and there are fewer constraints on the quality of sorting, that is, the correspondence between job characteristics and worker skills is less imperative. In this configuration the productivity level of each sorting is suboptimal. Conversely, when economic conditions worsen, lesser-quality assortments are the first to be eliminated and good assortments ensuring decent productivity levels are least affected
The competition between employers and workers
In addition to these first parameters – the quantity and quality of jobs and of workers – there is a salary dimension. To address it, the model assumes that employers are in a strong position when negotiating with the unemployed. The former offer the latter their reserve salary, that is, the minimum amount that the unemployed are ready to accept to work. Employers then pocket all of what is called the salary rent (i.e. the difference between what the pair produces and the unemployment value). But the unemployed are not the only ones receiving job offers: also affected are workers who already have a job and can be poached. When workers receive alternative offers, they will take them if their employers cannot offer anything better than the alternative salary. These workers can also take advantage of the situation to negotiate a salary increase. Thus, thanks to competition, employers are forced to share the rent at some point.
The competition between firms to hire the best-adapted workers therefore has two effects. First, it promotes labor reallocations and contributes to general growth. Second, it allows for more equal rent sharing through salary increases.
The model also looks at the expectations of actors who know that competitive pressure varies with economic conditions. While nobody can predict events that are by nature random – aggregate productivity shocks, layoff shocks, alternative job offers, etc. – workers and owners know the probability of these events occurring. The model therefore posits that their expectations are rational. It is a hypothesis that carries a lot of weight in the model but that is audacious in its strength, but very little is known about the way in which agents form expectations.
The model is tested on a significant series of data
Briefly, here is the story that the model is trying to tell – or rather “the possible stories”, since the model contains so many parameters to calibrate (15 in total). These parameters are set in such a way that the economy that the model is simulating closely reproduces various American statistics (26 in total) like temporal averages and volatilities in unemployment rates over length of unemployment, employment-to-unemployment and employment-to-employment transition rates, number of vacancies, correlations between these statistics, etc.
The model’s adjustment to the data has proven to be excellent. One process alone – the aggregate productivity shock – has proven to suffice to predict unemployment dynamics, the number of jobs created, the rate of return to employment for the unemployed, the rate of employment destruction and the level of employment-to-employment mobility.
Good economic conditions are not conducive to the best sorting
It is therefore possible to predict that in the aftermath of a recession employment increases as the least qualified workers find work again thanks to the reappearance of low-productivity jobs. Conversely, when the economy enters a recessive phase, the least qualified workers are the first ones to be fired, especially those who had found high-value-added jobs, which then most often go to the most competent workers.
Jean-Marc Robin is head of Sciences Po’s department of economics. He is also a professor at the University College of London, and a member of the Centre for Microdata Methods and Practice, of the Institute for Fiscal Studies and of the Econometric Society. His research in micro-econometrics particularly focus on
labor micro-econometrics and sorting. In 2010 he was awarded an Advanced ERC grant: WASP – Wage Dynamics, Sorting Patterns in Labour Markets and Policy Evaluation (FR)
CONDITIONS DE REPUBLICATION
Vous pouvez republier cet article en ligne ou sur papier sous notre licence Creative Commons. Vous ne pouvez pas modifier ou raccourcir le texte, vous devez attribuer l'article à Cogito et vous devez inclure le nom de l'auteur dans votre republication.
Si vous avez des questions, veuillez envoyer un courriel à email@example.com